{"id":{"repo_id":"msu","oai_identifier":"oai:d.lib.msu.edu:etd_52992"},"canonical_url":"https://search.ndltd.org/etd/msu/oai:d.lib.msu.edu:etd_52992","repository":{"repo_id":"msu","name":"Michigan State University","base_url":"https://d.lib.msu.edu/oai"},"display":{"title":"STATISTICAL METHODS DEVELOPMENT FOR MULTI-OMICS STUDIES AND GENOMIC PREDICTION","abstract":"Thesis (Ph.D.)--Michigan State University. Biostatistics - Doctor of Philosophy, 2026","abstract_html":"Thesis (Ph.D.)--Michigan State University. Biostatistics - Doctor of Philosophy, 2026","abstract_has_math":false,"creators":["Wu, Hao"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":["de los Campos, Gustavo","Vazquez, Ana Ines","Cui, Yuehua","Liang, Xiaoyu"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-11-18T18:59:34Z","subjects":["Biostatistics","Genetics","Statistics","Biometry"],"languages":["eng"],"rights":["Attribution 4.0 International"],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:52992","local:Wu_grad.msu_0128D_21992"],"render_values":[{"text":"etd:52992","href":null,"code":true},{"text":"local:Wu_grad.msu_0128D_21992","href":null,"code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://doi.org/doi:10.25335/1mm7-7c49","outbound_label":"DOI","outbound_source":"dc:identifier","record_url":"https://doi.org/doi:10.25335/1mm7-7c49","record_label":"Repository record","record_kind":"doi","record_source":"dc:identifier","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://d.lib.msu.edu/oai?verb=GetRecord&metadataPrefix=oai_dc&identifier=oai%3Ad.lib.msu.edu%3Aetd_52992","prefix":"oai_dc"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["de los Campos, Gustavo","Vazquez, Ana Ines","Cui, Yuehua","Liang, Xiaoyu"]},{"key":"dc:creator","label":"Author","values":["Wu, Hao"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2026"]},{"key":"dc:relation","label":"Dc Relation","values":["Electronic Theses & Dissertations"]},{"key":"dc:type","label":"Dc Type","values":["text","Electronic dissertations"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Biostatistics","Genetics","Statistics","Biometry"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["Attribution 4.0 International"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:52992","local:Wu_grad.msu_0128D_21992","https://doi.org/doi:10.25335/1mm7-7c49"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Thesis (Ph.D.)--Michigan State University. Biostatistics - Doctor of Philosophy, 2026","The era of big data has created unprecedented opportunities to uncover insights across a wide range of scientific disciplines. In omics research, advances in sequencing technology have enabled comprehensive characterization of biological systems such as the microbiome and metabolome, deepening our understanding of human health and diseases. In genetics, the rapid growth of genome-wide association studies has facilitated the identification of genetic variants associated with complex human traits and informed the prediction of such phenotypes. Despite these advances, important methodological challenges remain, particularly in modeling complex multi-omics data and improving the portability and scalability of omics (including genomic prediction) models across populations, environments, and domains. This dissertation develops statistical methods to address these challenges and is organized into three complementary studies.Chapter 1 develops a statistical framework to characterize subject-specific longitudinal dynamics in microbiome-metabolome data. While prior studies have primarily focused on population-level temporal trends, less is known about heterogeneity across individuals and the extent to which microbial and metabolite trajectories are associated over time. To address this, we used mixed-effects models to capture both shared and subject-specific patterns in longitudinal profiles. Applying the proposed approach to data from the ABC Baby Study, we identified coordinated temporal dynamics between microbial taxa and metabolites during early childhood. In particular, we detected numerous microbe-metabolite pairs exhibiting synchronized presence/absence patterns over time. Enrichment analyses revealed that specific microbial families and metabolite groups disproportionately contribute to these coordinated changes. These findings provide new insights into the joint development of the microbiome and metabolome and highlight substantial inter-individual variability in early-life biological processes. The results from this chapter were published in Communications Biology (Wu et al., 2024).Chapter 2 addresses a major limitation of polygenic score (PGS) prediction: reduced predictive performance in non-European populations due to the imbalance of available training data. To mitigate this issue, we introduced Transfer Learning (TL) methods for genomic prediction and presented GPTL, an R package implementing three complementary approaches based on informative shrinkage and Bayesian modeling. These methods incorporate information from well-represented populations by using estimated SNP (single nucleotide polymorphism) effects as prior information for a TL algorithm that is run on data from the target population. This approach allows target-population-specific estimates and borrowing of information between populations. We evaluated these approaches using both simulated data and large-scale biobank datasets, including UK Biobank and All of Us, under a variety of genetic architectures and cross-ancestry scenarios. Results showed that TL-based PGS consistently outperformed single-ancestry models and, in many cases, achieved performance comparable to or exceeding multi-ancestry methods. In addition, the framework accommodates both individual-level data and summary statistics, providing flexibility for applications where data sharing is limited. The results from this chapter were published in Nature Communications (Wu et al., 2026).Chapter 3 extends the TL framework to genomic prediction in plant and animal breeding, where data are collected across heterogeneous contexts such as sub-populations, environments, and breeding cycles. In these settings, both statistical and computational challenges arise due to heterogeneity in genetic effects and the increasing scale of the available data. We provided a systematic evaluation of TL methods relative to competitive benchmarks across a range of realistic prediction settings. Across multiple case studies, TL achieves predictive performance comparable to or better than the strongest benchmarks, particularly when there is substantial effect heterogeneity between source and target contexts (e.g., across environments or across sub-populations) and the target data sample size is limited. In addition, TL offers substantial computational advantages in sequential settings by avoiding repeated estimation on cumulative data. The results also highlight the potential of TL to facilitate information sharing across datasets when individual-level data cannot be exchanged, as knowledge can be transferred through summary statistics such as estimated marker effects.Overall, this dissertation develops statistical methods for analyzing multi-omics data and improving genomic prediction in heterogeneous settings, addressing both inference and prediction challenges in modern biological studies.","Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).","Includes bibliographical references."]},{"key":"dc:format","label":"Dc Format","values":["application/pdf","1 online resource (127 p.) : ill."]},{"key":"dc:title","label":"Title","values":["STATISTICAL METHODS DEVELOPMENT FOR MULTI-OMICS STUDIES AND GENOMIC PREDICTION"]}]}],"canonical_facts":{"dc:contributor":["de los Campos, Gustavo","Vazquez, Ana Ines","Cui, Yuehua","Liang, Xiaoyu"],"dc:creator":["Wu, Hao"],"dc:date":["2026"],"dc:description":["Thesis (Ph.D.)--Michigan State University. Biostatistics - Doctor of Philosophy, 2026","The era of big data has created unprecedented opportunities to uncover insights across a wide range of scientific disciplines. In omics research, advances in sequencing technology have enabled comprehensive characterization of biological systems such as the microbiome and metabolome, deepening our understanding of human health and diseases. In genetics, the rapid growth of genome-wide association studies has facilitated the identification of genetic variants associated with complex human traits and informed the prediction of such phenotypes. Despite these advances, important methodological challenges remain, particularly in modeling complex multi-omics data and improving the portability and scalability of omics (including genomic prediction) models across populations, environments, and domains. This dissertation develops statistical methods to address these challenges and is organized into three complementary studies.Chapter 1 develops a statistical framework to characterize subject-specific longitudinal dynamics in microbiome-metabolome data. While prior studies have primarily focused on population-level temporal trends, less is known about heterogeneity across individuals and the extent to which microbial and metabolite trajectories are associated over time. To address this, we used mixed-effects models to capture both shared and subject-specific patterns in longitudinal profiles. Applying the proposed approach to data from the ABC Baby Study, we identified coordinated temporal dynamics between microbial taxa and metabolites during early childhood. In particular, we detected numerous microbe-metabolite pairs exhibiting synchronized presence/absence patterns over time. Enrichment analyses revealed that specific microbial families and metabolite groups disproportionately contribute to these coordinated changes. These findings provide new insights into the joint development of the microbiome and metabolome and highlight substantial inter-individual variability in early-life biological processes. The results from this chapter were published in Communications Biology (Wu et al., 2024).Chapter 2 addresses a major limitation of polygenic score (PGS) prediction: reduced predictive performance in non-European populations due to the imbalance of available training data. To mitigate this issue, we introduced Transfer Learning (TL) methods for genomic prediction and presented GPTL, an R package implementing three complementary approaches based on informative shrinkage and Bayesian modeling. These methods incorporate information from well-represented populations by using estimated SNP (single nucleotide polymorphism) effects as prior information for a TL algorithm that is run on data from the target population. This approach allows target-population-specific estimates and borrowing of information between populations. We evaluated these approaches using both simulated data and large-scale biobank datasets, including UK Biobank and All of Us, under a variety of genetic architectures and cross-ancestry scenarios. Results showed that TL-based PGS consistently outperformed single-ancestry models and, in many cases, achieved performance comparable to or exceeding multi-ancestry methods. In addition, the framework accommodates both individual-level data and summary statistics, providing flexibility for applications where data sharing is limited. The results from this chapter were published in Nature Communications (Wu et al., 2026).Chapter 3 extends the TL framework to genomic prediction in plant and animal breeding, where data are collected across heterogeneous contexts such as sub-populations, environments, and breeding cycles. In these settings, both statistical and computational challenges arise due to heterogeneity in genetic effects and the increasing scale of the available data. We provided a systematic evaluation of TL methods relative to competitive benchmarks across a range of realistic prediction settings. Across multiple case studies, TL achieves predictive performance comparable to or better than the strongest benchmarks, particularly when there is substantial effect heterogeneity between source and target contexts (e.g., across environments or across sub-populations) and the target data sample size is limited. In addition, TL offers substantial computational advantages in sequential settings by avoiding repeated estimation on cumulative data. The results also highlight the potential of TL to facilitate information sharing across datasets when individual-level data cannot be exchanged, as knowledge can be transferred through summary statistics such as estimated marker effects.Overall, this dissertation develops statistical methods for analyzing multi-omics data and improving genomic prediction in heterogeneous settings, addressing both inference and prediction challenges in modern biological studies.","Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).","Includes bibliographical references."],"dc:format":["application/pdf","1 online resource (127 p.) : ill."],"dc:identifier":["etd:52992","local:Wu_grad.msu_0128D_21992","https://doi.org/doi:10.25335/1mm7-7c49"],"dc:language":["eng"],"dc:relation":["Electronic Theses & Dissertations"],"dc:rights":["Attribution 4.0 International"],"dc:subject":["Biostatistics","Genetics","Statistics","Biometry"],"dc:title":["STATISTICAL METHODS DEVELOPMENT FOR MULTI-OMICS STUDIES AND GENOMIC PREDICTION"],"dc:type":["text","Electronic dissertations"]},"updated_at":"2026-11-18T18:59:34Z"}